大型实验仪器设备异常运行状态辨识方法研究

Research on identification method of abnormal operation state of large-scale experimental instruments and equipment

以往辨识方法无法正确选择有效特征,导致辨识精准度较低,为了改善该问题,提出了基于学习机器支持向量机的大型实验仪器设备异常运行状态辨识方法。根据时域统计均值、方差、均方根值特征量,消除次要参量干扰,提取特征集中最优参量。通过核函数将最优参量映射到核空间中,形成映射向量。依据识别函数在特征空间中构造时频平面,采用多分类支持向量机对映射向量进行训练与自学习,将训练与学习结果作为设备异常状态的辨识标准。确定异常状态辨识因子,结合辨识标准,完成大型实验仪器设备异常运行状态辨识。通过实验对比结果可知,该方法辨识精准度较高,为大型实验仪器设备安全使用提供依据。

In order to improve the problem,a new method based on learning machine support vector machine(LMSVM)for identifying abnormal operating state of large-scale experimental equipment is proposed. According to the statistical mean,variance and root mean square eigenvalues in time domain,the interference of secondary parameters is eliminated and the optimal parameters in feature set are extracted. The optimal parameters are mapped into the kernel space by the kernel function,and the mapping vectors are formed. According to the recognition function,the time-frequency plane is constructed in the feature space,and the mapping vector is trained and self-learnt by using multi-class support vector machine. The training and learning results are taken as the identification criteria of the abnormal state of the equipment. The abnormal state identification factor is determined,and the abnormal operation state identification of large-scale experimental equipment is completed by combining the identification criteria. The comparison results show that the identification accuracy of this method is high,which provides a basis for the safe use of large-scale experimental equipment.